# US 2023/0276261 A1 · Customization feature vector feedback for machine learning components in wireless communication

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2023/0276261 A1
- Document ID: US20230276261A1
- Status: Published application
- Kind: A1
- Application number: 18/006,693
- Filed: 2021-07-30
- Published: 2023-08-31
- Earliest claimed: 2020-09-11
- Assignees: QUALCOMM Incorporated
- Applicants: QUALCOMM Incorporated
- Inventors: Alexandros MANOLAKOS; June NAMGOONG; Naga BHUSHAN; Taesang YOO; Pavan Kumar VITTHALADEVUNI; Jay Kumar SUNDARARAJAN; Krishna Kiran MUKKAVILLI; Hwan Joon KWON; Tingfang JI
- Web page: https://patentgrep.com/patent/US20230276261A1

## CPC

- G06N 3/088: Non-supervised learning, e.g. competitive learning
- H04W 24/02: Arrangements for optimising operational condition
- G06N 3/045: Combinations of networks
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/0495: Quantised networks; Sparse networks; Compressed networks
- G06N 3/063: using electronic means
- G06N 3/098: Distributed learning, e.g. federated learning
- H04W 24/10: Scheduling measurement reports {; Arrangements for measurement reports}
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- H04W: WIRELESS COMMUNICATION NETWORKS (broadcast communication H04H; communication systems using wireless links for non-selective communication, e.g. wireless extensions H04M1/72)

## Abstract

Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a client may receive a customization feature vector feedback configuration associated with a reporting procedure for reporting updates corresponding to at least one customization feature vector that is based at least in part on one or more features associated with an environment of the client. The client may determine an update corresponding to the at least one customization feature vector using a machine learning component. The client may transmit the update based at least in part on the customization feature vector feedback configuration. Numerous other aspects are provided.

## Claims

### Claim 1

An apparatus for wireless communication at a client, comprising:
a memory; and
one or more processors, coupled to the memory, configured to:
receive a customization feature vector feedback configuration associated with a reporting procedure for reporting updates corresponding to at least one customization feature vector that is based at least in part on one or more features associated with an environment of the client;
determine an update corresponding to the at least one customization feature vector using a machine learning component; and
transmit the update based at least in part on the customization feature vector feedback configuration.

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein the update comprises a locally updated customization feature vector.

### Claim 3 (depends on 1)

The apparatus of claim 1, wherein the at least one customization feature vector comprises a plurality of customization feature vectors.

### Claim 4 (depends on 3)

The apparatus of claim 3, wherein the plurality of customization feature vectors corresponds to a plurality of communication parameters.

### Claim 5 (depends on 4)

The apparatus of claim 4, wherein the plurality of communication parameters indicate at least one of:
a component carrier,
a bandwidth part,
a band,
a band combination,
a frequency range,
a use case,
a neural network identifier, or
a combination thereof.

### Claim 6 (depends on 5)

The apparatus of claim 5, wherein the use case comprises at least one of:
a channel state information derivation,
a positioning measurement derivation,
demodulation of a data channel, or
a combination thereof.

### Claim 7 (depends on 1)

The apparatus of claim 1, wherein the one or more processors are further configured to:
transmit a capability report that indicates a number of customization feature vectors that the client can support;
wherein the customization feature vector configuration is based at least in part on the capability report.

### Claim 8 (depends on 1)

The apparatus of claim 1, wherein the customization feature vector configuration indicates a transmission parameter that indicates a quantity of scheduled uplink resources that can be used to carry one or more portions of the update.

### Claim 9 (depends on 1)

The apparatus of claim 1, wherein determining the update comprises determining a locally updated customization feature vector, and the one or more processors, to transmit the update, are configured to:
quantize the locally updated customization feature vector to generate a quantized update; and
transmit the quantized update.

### Claim 10 (depends on 9)

The apparatus of claim 9, wherein the one or more processors are further configured to determine an indication of a maximum bit size corresponding to the quantized update,
wherein the quantized update comprises a variable bit number, and
wherein the variable bit number comprises a bit size that is less than or equal to the maximum bit size.

### Claim 11 (depends on 9)

The apparatus of claim 9, wherein the quantized update comprises a non-uniform quantization of one or more elements of the locally updated customization feature vector, wherein the one or more processors are further configured to receive an indication of a non-uniform quantization scheme, wherein the non-uniform quantization is based at least in part on the indication of the non-uniform quantization scheme, wherein the non-uniform quantization scheme is carried in at least one of:
the customization feature vector configuration,
a radio resource control message,
a downlink control information transmission,
a medium access control (MAC) control element, or
a combination thereof.

### Claim 12 (depends on 11)

The apparatus of claim 11, wherein the one or more processors are further configured to transmit a non-uniform quantization suggestion,
wherein the non-uniform quantization scheme is based at least in part on the non-uniform quantization suggestion.

### Claim 13 (depends on 1)

The apparatus of claim 1, wherein the one or more processors are further configured to:
receive a latent vector configuration that indicates a periodic reporting procedure for reporting updates corresponding to at least one latent vector;
determine an update corresponding to the at least one latent vector using an additional machine learning component; and
transmit the update corresponding to the at least one latent vector based at least in part on the latent vector configuration.

### Claim 14 (depends on 13)

The apparatus of claim 13, wherein the latent vector configuration indicates a first periodicity associated with reporting updates corresponding to the at least one latent vector, wherein the first periodicity comprises an amount of time between consecutive transmissions of the updates corresponding to the at least one latent vector; and
wherein the customization feature vector configuration indicates a second periodicity associated with reporting the updates corresponding to the at least one customization feature vector, wherein the second periodicity comprises an amount of time between consecutive transmissions of the updates corresponding to the at least one customization feature vector.

### Claim 15 (depends on 13)

The apparatus of claim 13, wherein the latent vector configuration indicates a first periodicity associated with reporting updates corresponding to the at least one latent vector, wherein the first periodicity comprises an amount of time between consecutive transmissions of the updates corresponding to the at least one latent vector; and
wherein the customization feature vector configuration indicates a scaling factor for determining a second periodicity relative to the first periodicity, wherein the second periodicity comprises an amount of time between consecutive transmissions of the updates corresponding to the at least one customization feature vector.

### Claim 16 (depends on 15)

The apparatus of claim 15, wherein the one or more processors are further configured to:
identify a collision between a scheduled transmission of an update corresponding to the at least one latent vector and a scheduled transmission of an update corresponding to the at least one customization feature vector; and
transmit the scheduled transmission of the update corresponding to the at least one customization feature vector.

### Claim 17 (depends on 15)

The apparatus of claim 15, wherein the at least one customization feature vector comprises a plurality of customization feature vectors, and
wherein the customization feature vector configuration indicates an ordering associated with transmitting a plurality of updates corresponding to the plurality of customization feature vectors.

### Claim 18 (depends on 17)

The apparatus of claim 17, wherein the one or more processors are further configured to:
identify a collision between a scheduled transmission of the plurality of updates corresponding to the plurality of customization feature vectors and an additional transmission; and
refrain from transmitting an end portion of a report comprising the plurality of updates corresponding to the plurality of customization feature vectors based at least in part on identifying the collision.

### Claim 19 (depends on 17)

The apparatus of claim 17, wherein the one or more processors are further configured to:
receive a resource allocation associated with a scheduled transmission of the plurality of updates corresponding to the plurality of customization feature vectors and an additional transmission;
determine that the resource allocation is insufficient to transmit a report comprising the plurality of updates corresponding to the plurality of customization feature vectors; and
refrain from transmitting an end portion of the report based at least in part on determining that the resource allocation is insufficient to transmit the report.

### Claim 20 (depends on 15)

The apparatus of claim 15, wherein the at least one customization feature vector comprises a plurality of customization feature vectors and wherein the at least one latent vector comprises a plurality of latent vectors, and
wherein the transmitting the update comprises:
transmit, according to an ordering, a first plurality of updates, wherein the first plurality of updates correspond to the plurality of customization feature vectors; and
transmit, according to the ordering, a second plurality of updates, wherein the second plurality of updates correspond to the plurality of latent vectors.

### Claim 21 (depends on 1)

The apparatus of claim 1, wherein the one or more processors are further configured to:
receive a latent vector configuration that indicates a dynamic reporting procedure for reporting updates corresponding to at least one latent vector;
determine an update corresponding to the at least one latent vector using an additional machine learning component; and
transmit the update corresponding to the at least one latent vector based at least in part on the latent vector configuration.

### Claim 22 (depends on 21)

The apparatus of claim 21, wherein the customization feature vector configuration comprises an indication to transmit the update corresponding to the at least one customization feature vector based at least in part on receiving a latent vector feedback request.

### Claim 23 (depends on 22)

The apparatus of claim 22, wherein the latent vector feedback request comprises a trigger that triggers transmission of an update corresponding to the at least one latent vector, and
wherein the trigger further triggers transmission of the update corresponding to the at least one customization feature vector.

### Claim 24 (depends on 21)

The apparatus of claim 21, wherein the one or more processors are further configured to transmit a feedback resource request, wherein the feedback resource request comprises a request for:
a first set of resources for transmitting the update corresponding to the at least one latent vector,
a second set of resources for transmitting the update corresponding to the at least one customization feature vector, or
a combination thereof.

### Claim 25 (depends on 21)

The apparatus of claim 21, wherein the one or more processors are further configured to:
determine an occurrence of an update reporting trigger event; and
transmit, based at last on determining the occurrence of the update reporting trigger event, at least one of the update corresponding to the at least one customization feature vector or the update corresponding to the at least one latent vector.

### Claim 26 (depends on 21)

The apparatus of claim 21, wherein the one or more processors, to determine the update corresponding to the at least one latent vector, are configured to determine a locally updated latent vector based at least in part on a previously reported customization feature vector.

### Claim 27 (depends on 21)

The apparatus of claim 21, wherein the one or more processors, to determine the update corresponding to the at least one latent vector, are configured to determine a locally updated latent vector based at least in part on a default customization feature vector.

### Claim 28

An apparatus for wireless communication at a server, comprising:
a memory; and
one or more processors, coupled to the memory, configured to:
transmit a customization feature vector feedback configuration associated with a reporting procedure for reporting updates corresponding to at least one customization feature vector that is based at least in part on a machine learning component and one or more features associated with an environment of a client; and
receive an update corresponding to the at least one customization feature vector based at least in part on the configuration.

### Claim 29

A method of wireless communication performed by a client, comprising:
receiving a customization feature vector feedback configuration associated with a reporting procedure for reporting updates corresponding to at least one customization feature vector that is based at least in part on one or more features associated with an environment of the client;
determining an update corresponding to the at least one customization feature vector using a machine learning component; and
transmitting the update based at least in part on the customization feature vector feedback configuration.

### Claim 30

A method of wireless communication performed by a server, comprising:
transmitting a customization feature vector feedback configuration associated with a reporting procedure for reporting updates corresponding to at least one customization feature vector that is based at least in part on a machine learning component and one or more features associated with an environment of a client; and
receiving an update corresponding to the at least one customization feature vector based at least in part on the configuration.
